From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from yw-out-2324.google.com (yw-out-2324.google.com [74.125.46.29]) by ozlabs.org (Postfix) with ESMTP id 415AEDDF87 for ; Fri, 8 May 2009 00:09:23 +1000 (EST) Received: by yw-out-2324.google.com with SMTP id 2so428297ywt.39 for ; Thu, 07 May 2009 07:09:21 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <20090506224134.83F1583420E8@gemini.denx.de> References: <1241640919-4650-1-git-send-email-wd@denx.de> <1241640919-4650-3-git-send-email-wd@denx.de> <20090506220119.7322783420E8@gemini.denx.de> <20090506224134.83F1583420E8@gemini.denx.de> From: Grant Likely Date: Thu, 7 May 2009 08:09:01 -0600 Message-ID: Subject: Re: [PATCH 02/12] fs_enet: Add MPC5121 FEC support. To: Wolfgang Denk Content-Type: text/plain; charset=ISO-8859-1 Cc: Piotr Ziecik , Detlev Zundel , netdev@vger.kernel.org, linuxppc-dev@ozlabs.org, John Rigby List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wed, May 6, 2009 at 4:41 PM, Wolfgang Denk wrote: > Dear Grant, > > In message = you wrote: >> >> > Agreed that it's ugly, but duplicatio9ng the code would have been even >> > worse. I don't think that it has multiplatform - at least not as long >> > as you don't ask for one image that runs on 83xx and on 512x. >> >> Actually, I *am* asking for one image that runs on 83xx, 52xx and >> 521x. =A0I already can and do build and test a single image which boots >> on all my 52xx boards, on my 8349 board, and on my G4 Mac. > > He. I was afraid you'd say that ;-) > > In this case I need a helping hand as I can't figure out how to make > this work. Any suggestions? Hmmm, it is rather ugly because the layout of fec_t is so different. Easiest solution would be to duplicate the driver in its entirety, but as you say it results in a lot of duplicate code. It might be the lesser of many weevils though. Second easiest would be to factor out all the common code in the driver into a separate .c file that gets included by a 'wrapper' .c file for each config which first includes the correct definition of fec_t. This approach solves the duplicate code problem, but it also fell out of the ugly tree and hit every branch on the way down. ie: the wrappers would look something like this: fs_enet_cpm1.c: #include #include "fs_enet_main.c" fs_enet_cpm2.c: #include #include "fs_enet_main.c" fs_enet_512x.c: #include #include "fs_enet_main.c" And then the makefile would be something along the lines of: obj-${CONFIG_FS_ENET_CPM1_ +=3D fs_enet_cpm1.o obj-${CONFIG_FS_ENET_CPM2_ +=3D fs_enet_cpm2.o obj-${CONFIG_FS_ENET_MPC512x_ +=3D fs_enet_512x.o A brief look at the driver suggests that access to the fec_t structure is restricted to the fec-mii.c and mac-fec.c files. It might be appropriate to duplicate just those files and do some form of fs_enet_ops to select between them. While on the topic, it looks to me like the driver could really use some refactoring love. Having multiple definitions of "fec_t" is confusing and potentially lead to hard to find bugs if the wrong header gets included by anyone. I'd like to see all the fec specific stuff in arch/powerpc/include/asm moved into drivers/net/fs_enet and renamed to reflect the hardware it is associate with. Stuff like "fec_t" is far to generic, not to mention that typedefs are discouraged now. Regardless, I feel your pain. It is not a pretty situation. g. --=20 Grant Likely, B.Sc., P.Eng. Secret Lab Technologies Ltd.